3 ms·

I’m really curious as to how on Earth did Google fumble the ball on this stuff? It’s hilarious that their search offering has been so bad for so long that the

I’m really curious as to how on Earth did Google fumble the ball on this stuff? It’s hilarious that their search offering has been so bad for so long that the new Big Idea in Search is “What if you searched something and the returned result wasn’t garbage?”

TaylorAlexander 4y ago
I don't see it as fumbling the ball. The technology is brand new and Google just yesterday announced they are currently testing bringing this to chat too. Just like Apple does not release folding phones and rear screens first, the major player does not need to be the first mover. OpenAI has grabbed all the publicity but this entire time Google research has been publishing their own work in the field and they have shown it off at the annual google developer conference. Now the day after Google announced that they are beginning large scale testing of this feature in search for an imminent wide release, Microsoft has released their version. There is no need for Google to be the absolute first. What is most important is that they keep the Google brand looking great and they release the feature when it is ready.
- zone411 4y ago
  They pretty much admitted that they did fumble the ball: https://www.nytimes.com/2022/12/21/technology/ai-chatgpt-google-search.html https://www.nytimes.com/2022/12/21/technology/ai-chatgpt-goo.... But I agree that Google's research is #1 and they will come back strong.
- speedgoose 4y ago
  Meanwhile ChatGPT had 100 million users and 13 million daily users in January. They accumulated a lot of data and while I’m sure the Google engineers are very competent and fast, their large scale testing may not produce as much useful data. I think Google will manage in the long term, but for the current race it looks difficult.
  - TaylorAlexander 4y ago
    This all assumes it’s even a good idea. Google is always extremely careful with their brand. We harp on them for canceling projects but to most people google is a search engine. And yes the quality of results has steadily declined for a long time, but deploying a large language model in google search has the potential to cause serious negative outcomes unlike slowly degrading search results. I really don’t see it as a race, which presumes that moving first is an advantage. OpenAI is building their brand and wants to be known as an AI company, so it makes sense to release early versions of their system. Google is a well established company that can easily deploy this technology when they think it’s ready and eclipse OpenAI in users overnight. But even then it’s not about having the most users, it’s about building a new platform for the next 10+ years of search. Google is already the default. They can take their time and still do just fine.
dougmwne 4y ago
It’s not a fumble, more like a fundamental limit of the public corporation. Google has an established cash flow that relies on search and people clicking through to websites full of more ads. Once Google perfected that revenue model, they were pretty much locked in. They could explore new revenue streams in new areas like cars or speakers, but investors would howl for blood if they tried to reinvent the search experience at the cost of revenue growth. These LLMs change the revenue model too much. Imagine how many ads a person sees over the course of researching the best TV to buy. Clicking and clicking on dozens of pages and seeing hundreds of ads. In the Bing demo, they asked for the best gaming TV and boom, there was the gaming TV I just bought after doing the research myself the long way. No clicks, no revenue.
- braingenious 4y ago
  I don’t really understand this line of reasoning. >In the Bing demo, they asked for the best gaming TV and boom, there was the gaming TV I just bought after doing the research myself the long way. No clicks, no revenue. In this example “no clicks, no revenue” applies more to Google than it does Bing. Bing maybe didn’t make any money off that particular search but they did demonstrate that they have a better product, increasing the chances that someone will use Bing instead of Google. Bing didn’t get the click revenue, but Google didn’t even get the search in the first place. The way forward for Google is to… continue to maximize dollars per click? The exact same behavior that’s made their search laughably bad for many people?
  - dougmwne 4y ago
    The reasoning is: Google should continue to maximize dollars per click. They have an established revenue model and should not try to disrupt themselves. Short term profits over long term profits are favored by Wall Street. Microsoft has a tiny portion of the search and ads market. They have no substantial revenue to protect. If they try to disrupt search and succeed, but kill the click through ratio and number of searches per user in the process, that’s no problem for them. They can steal away Google’s users first and worry about making money off it later. They might even make search less profitable in the long run and something that offered as a loss leader to get users into the rest of their ecosystem.
HDThoreaun 4y ago
If google goes from profiting $.001 to $.0005 per search they'd face an existential threat. That may well be the reality of chatbot based search platforms. Microsoft is happy to hoover up those searches at even $.00005 profit a hit, but for google that would mean losing $50 billion a year in profit.
seydor 4y ago
To their credit, they did most of the research for this, transformers + reinforcement learning. But maybe they could not see a way that this could be used to lock in both advertisers and publishers in their own 'free' marketplace

TaylorAlexander 4y ago

I don't see it as fumbling the ball. The technology is brand new and Google just yesterday announced they are currently testing bringing this to chat too. Just like Apple does not release folding phones and rear screens first, the major player does not need to be the first mover. OpenAI has grabbed all the publicity but this entire time Google research has been publishing their own work in the field and they have shown it off at the annual google developer conference. Now the day after Google announced that they are beginning large scale testing of this feature in search for an imminent wide release, Microsoft has released their version. There is no need for Google to be the absolute first. What is most important is that they keep the Google brand looking great and they release the feature when it is ready.

zone411 4y ago
They pretty much admitted that they did fumble the ball: https://www.nytimes.com/2022/12/21/technology/ai-chatgpt-google-search.html https://www.nytimes.com/2022/12/21/technology/ai-chatgpt-goo.... But I agree that Google's research is #1 and they will come back strong.
speedgoose 4y ago
Meanwhile ChatGPT had 100 million users and 13 million daily users in January. They accumulated a lot of data and while I’m sure the Google engineers are very competent and fast, their large scale testing may not produce as much useful data. I think Google will manage in the long term, but for the current race it looks difficult.
- TaylorAlexander 4y ago
  This all assumes it’s even a good idea. Google is always extremely careful with their brand. We harp on them for canceling projects but to most people google is a search engine. And yes the quality of results has steadily declined for a long time, but deploying a large language model in google search has the potential to cause serious negative outcomes unlike slowly degrading search results. I really don’t see it as a race, which presumes that moving first is an advantage. OpenAI is building their brand and wants to be known as an AI company, so it makes sense to release early versions of their system. Google is a well established company that can easily deploy this technology when they think it’s ready and eclipse OpenAI in users overnight. But even then it’s not about having the most users, it’s about building a new platform for the next 10+ years of search. Google is already the default. They can take their time and still do just fine.

dougmwne 4y ago

It’s not a fumble, more like a fundamental limit of the public corporation. Google has an established cash flow that relies on search and people clicking through to websites full of more ads. Once Google perfected that revenue model, they were pretty much locked in. They could explore new revenue streams in new areas like cars or speakers, but investors would howl for blood if they tried to reinvent the search experience at the cost of revenue growth. These LLMs change the revenue model too much. Imagine how many ads a person sees over the course of researching the best TV to buy. Clicking and clicking on dozens of pages and seeing hundreds of ads. In the Bing demo, they asked for the best gaming TV and boom, there was the gaming TV I just bought after doing the research myself the long way. No clicks, no revenue.

braingenious 4y ago
I don’t really understand this line of reasoning. >In the Bing demo, they asked for the best gaming TV and boom, there was the gaming TV I just bought after doing the research myself the long way. No clicks, no revenue. In this example “no clicks, no revenue” applies more to Google than it does Bing. Bing maybe didn’t make any money off that particular search but they did demonstrate that they have a better product, increasing the chances that someone will use Bing instead of Google. Bing didn’t get the click revenue, but Google didn’t even get the search in the first place. The way forward for Google is to… continue to maximize dollars per click? The exact same behavior that’s made their search laughably bad for many people?
- dougmwne 4y ago
  The reasoning is: Google should continue to maximize dollars per click. They have an established revenue model and should not try to disrupt themselves. Short term profits over long term profits are favored by Wall Street. Microsoft has a tiny portion of the search and ads market. They have no substantial revenue to protect. If they try to disrupt search and succeed, but kill the click through ratio and number of searches per user in the process, that’s no problem for them. They can steal away Google’s users first and worry about making money off it later. They might even make search less profitable in the long run and something that offered as a loss leader to get users into the rest of their ecosystem.

HDThoreaun 4y ago

If google goes from profiting $.001 to $.0005 per search they'd face an existential threat. That may well be the reality of chatbot based search platforms. Microsoft is happy to hoover up those searches at even $.00005 profit a hit, but for google that would mean losing $50 billion a year in profit.

seydor 4y ago

To their credit, they did most of the research for this, transformers + reinforcement learning. But maybe they could not see a way that this could be used to lock in both advertisers and publishers in their own 'free' marketplace